Author + information
- Received October 27, 2002
- Revision received December 4, 2002
- Accepted December 26, 2002
- Published online April 16, 2003.
- David A. Morrow, MD, MPH*,* (, )
- James A. de Lemos, MD†,
- Marc S. Sabatine, MD, MPH*,
- Sabina A. Murphy, MPH‡,
- Laura A. Demopoulos, MD§,
- Peter M. DiBattiste, MD§,
- Carolyn H. McCabe, BS*,
- C.Michael Gibson, MD, MS, FACC‡,
- Christopher P. Cannon, MD, FACC* and
- Eugene Braunwald, MD, FACC*
- ↵*Reprint requests and correspondence:
Dr. David A. Morrow, Cardiovascular Division, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusetts 02115, USA.
Objectives This study was designed to evaluate B-type natriuretic peptide (BNP) for risk assessment and clinical decision making over a range of cut points, alone and with cardiac troponin I (cTnI), in patients with non–ST-elevation acute coronary syndromes (ACS).
Background B-type natriuretic peptide holds promise for risk stratification. Additional evidence regarding optimal decision limits, use in combination with troponin, and use in targeting therapy is needed before acceptance into clinical use for ACS.
Methods We evaluated BNP at baseline in 1,676 patients with non–ST-elevation ACS randomized to early invasive versus conservative management.
Results Patients with elevated BNP (>80 pg/ml; n = 320) were at higher risk of death at seven days (2.5% vs. 0.7%, p = 0.006) and six months (8.4% vs. 1.8%, p < 0.0001). The association between BNP and mortality at six months (adjusted odds ratio [OR] 3.3; 95% confidence interval [CI] 1.7 to 6.3) was independent of important clinical predictors, including cTnI and congestive heart failure (CHF). Patients with elevated BNP had a fivefold higher risk of developing new CHF by 30 days (5.9% vs. 1.0%, p < 0.0001). B-type natriuretic peptide added prognostic information to cTnI, discriminating patients at higher mortality risk among those with negative (OR 6.9; 95% CI 1.9 to 25.8) and positive (OR 4.1; 95% CI 1.9 to 9.0) baseline cTnI results. No difference was observed in the effect of invasive versus conservative management when stratified by baseline levels of BNP (pinteraction≥ 0.6).
Conclusions Elevated BNP (>80 pg/ml) at presentation identifies patients with non–ST-elevation ACS who are at higher risk of death and CHF and adds incremental information to cTnI. Additional work is needed to identify therapies that may reduce the risk associated with increased BNP.
B-type natriuretic peptide (BNP) is a cardiac neurohormone that is synthesized in ventricular myocardium and released in response to increased ventricular wall stress (1–3). Its diverse actions include natriuresis, vasodilation, inhibition of the renin-angiotensin-aldosterone system, and inhibition of sympathetic nerve activity (4). First studied as a diagnostic and prognostic marker among patients with congestive heart failure (CHF) (5–7), BNP was subsequently found to predict outcomes in patients with acute myocardial infarction (MI) (8,9). Specifically, when measured between one and four days after presentation with transmural infarction, an elevated plasma concentration of BNP was associated with increased mortality risk, independently of left ventricular function (8,10,11). We recently extended these findings across the spectrum of patients with acute coronary syndromes (ACS), including those with unstable angina (UA), among whom elevated levels of BNP predicted a twofold to threefold higher risk of death by 10 months (12). Together, these data suggest that measurement of BNP in patients presenting with suspected ACS could be an important addition to our current tools for risk stratification.
Before considering BNP for routine clinical application in ACS, consistent findings from independent data sets are needed. Moreover, the optimal decision limits for prognostic evaluation must be defined, and careful assessment of the utility of BNP in combination with cardiac troponin is required. Lastly, the utility of BNP for directing therapeutic decision making requires testing. We thus evaluated the predictive capacity of BNP over a range of clinical thresholds, alone and in combination with cardiac troponin I (cTnI), among patients with non–ST-elevation ACS enrolled in the Treat Angina with Aggrastat and Determine Cost of Therapy with an Invasive or Conservative Strategy (TACTICS)-Thrombolysis In Myocardial Infarction (TIMI) 18 trial (13). We also assessed whether BNP is useful in identifying patients who will derive greater benefit from an early invasive management strategy.
1.1 Study population and treatment
A total of 2,220 patients were enrolled in TACTICS-TIMI 18 between December 18, 1997, and December 22, 1999. The study design and primary results have been described previously (13). Patients ≥18 years old were eligible for enrollment if they experienced UA within the preceding 24 h, were candidates for coronary revascularization, and had at least one of the following: ST segment depression (≥0.05 mV) or transient ST elevation, T wave (≥0.3 mV) inversion in ≥2 leads, elevated cardiac markers of necrosis, or documented coronary disease. Exclusion criteria included persistent ST-segment elevation, recent revascularization, factors associated with increased risk of bleeding, severe CHF or cardiogenic shock, important systemic disease, or serum creatinine >2.5 mg/dl.
Patients were treated with aspirin, intravenous unfractionated heparin, and tirofiban (Merck and Co., West Point, Pennsylvania) and were randomized to an early invasive or conservative strategy. Patients in the early invasive strategy were to undergo coronary angiography between 4 and 48 h after randomization, and revascularization when feasible based on coronary anatomy. Patients in the conservative strategy were treated medically and were to undergo coronary angiography and revascularization only if they manifested recurrent ischemia at rest or with provocative testing. The study protocol was approved by the institutional review boards of each participating hospital, and patients provided written informed consent.
1.2 End points
Patients were followed up for six months for the development of recurrent clinical events, including death from any cause, new or recurrent MI, rehospitalization for ACS, and new or worsening CHF. Each of these end points, except new/worsening CHF, were adjudicated by an independent clinical end points committee, blinded to treatment assignment. The index diagnosis was established by the investigator based on local electrocardiographic and laboratory data.
1.3 BNP testing
The protocol specified that blood samples be obtained at enrollment by trained study personnel in citrate-anticoagulated tubes and plasma isolated within 60 min of sample acquisition. Plasma samples were stored at −20°C or colder at the enrolling site until shipped to the TIMI Biomarker Core Laboratory, where they were maintained at −80°C. Aliquots were shipped frozen to Biosite Incorporated (San Diego, California), where they were thawed and analyzed using an established immunoassay (12)by personnel blinded to treatment allocation and clinical outcomes. The primary decision limit of 80 pg/ml was pre-specified based on our previous work with this assay (12).
Levels of cTnI were measured in the TIMI Biomarker Core Laboratory using the ACS:180 Chemiluminescence cTnI Immunoassay (Bayer Diagnostics, Tarrytown, New York) (14). The minimum detectable concentration is 0.03 ng/ml. A prognostic decision limit of 0.1 ng/ml was used for all analyses based on previous work with this assay (15).
1.4 Statistical methods
Plasma concentrations of BNP are described as the median and interquartile range. The baseline characteristics of patients with and of those without elevated levels of BNP were compared using the Wilcoxon Rank Sums test for continuous variables and the chi-squared test for categorical variables. Correlation coefficients reported between continuous variables are based on a non-parametric method (Spearman correlation). Analysis of findings from the TIMI Angiographic Core Laboratory were performed for the subset of patients participating in the TACTICS-TIMI 18 angiographic substudy. Data regarding the number of diseased coronary arteries were based on local interpretation in all patients undergoing angiography.
The univariate association between BNP and clinical outcomes, alone and stratified by cTnI, was evaluated using the chi-square test for dichotomized comparisons and the chi-square test for trend across BNP quartiles. The log-rank test was performed for the Kaplan-Meier probability estimates (Fig. 1). Using logistic regression, the relationship between BNP (dichotomized) and outcomes was adjusted for the effects of age, gender, diabetes, ST-segment depression on the presenting ECG, history of CHF, evidence of heart failure (HF) at presentation, and baseline cTnI result. Age was treated as a continuous variable. ST depression and baseline cTnI were dichotomized at thresholds of ≥0.05 mV and ≥0.1 ng/ml, respectively. Clinical follow-up through six months was complete among >99% of patients, supporting the use of logistic regression as opposed to Cox proportional hazard modeling. The primary cut point for BNP (80 pg/ml) was pre-specified based on our previous work with this assay (12). In addition, an exploratory analysis of the predictive capacity of BNP at thresholds across the range of 40 to 160 pg/ml was performed using logistic regression. The relative predictive value at each threshold was assessed using the chi-squared values and adjusted risk relationships from logistic regression.
Testing for heterogeneity in the effect of the invasive strategy between patients with and those without elevated levels of BNP was performed using logistic regression with terms for the main effects and for the interaction of BNP status with treatment allocation. Analyses were performed using STATA v7-intercooled (STATA Corp., College Station, Texas). A p value (two-tailed) <0.05 was considered statistically significant.
A total of 1,676 samples were available for determination of BNP levels at enrollment in TACTICS-TIMI 18. The baseline characteristics of those without available samples were similar to those included in this substudy (data not shown). Median (25th to 75th percentile) levels of BNP were 20 pg/ml (<5 to 54 pg/ml) and 38 pg/ml (10 to 81 pg/ml) in patients with index diagnoses of UA and non–ST-elevation myocardial infarction (NSTEMI), respectively. The plasma concentration of BNP exceeded 80 pg/ml in 25.2% (n = 153) of patients presenting with NSTEMI, 15.6% (n = 167) of patients diagnosed with UA, and 10.1% (n = 67) of those with baseline levels of cTnI <0.1 ng/ml. Elevated levels of BNP were detected among 13.6% (n = 135) of those with UA and no history or current evidence of HF.
Patients with elevated levels of BNP were older, more often female, and more likely to have a history of HF and diabetes (Table 1). At presentation, patients with elevated BNP more frequently had ST-segment depression and tachycardia (>100 beats/min). There was no association between elevated levels of BNP and a history of prior MI. A modest correlation between baseline levels of BNP and peak creatine kinase-MB (CK-MB) was evident (ρ = 0.27, p < 0.001). Patients with elevated BNP were more likely to have multi-vessel (≥2) coronary artery disease (74% vs. 60%, p < 0.0001). In the sub-group of patients with films submitted for interpretation by the TIMI Angiographic Core Laboratory (n = 416), similar rates of impaired epicardial flow (TIMI flow grade 0/1/2: 37.3% vs. 34.7%, p = 0.7) and myocardial perfusion (TIMI myocardial perfusion grade 0/1: 52.6% vs. 53.8%, p = 0.9) were evident in the culprit territory among patients with and among those without elevated BNP, respectively.
2.1 Prediction of clinical events
Patients with baseline levels of BNP >80 pg/ml were at significantly higher risk of death by 30 days after presentation (5.0% vs. 1.2%, p < 0.0001). The higher mortality risk associated with elevated BNP was evident as early as seven days (2.5% vs. 0.74%, p = 0.006) and persisted through six months of follow-up (p < 0.0001, Fig. 1A). The association between elevated levels of BNP and mortality at six months was independent of other important clinical predictors available at presentation, including age, gender, diabetes, ST-segment depression, history of CHF, CHF at presentation, and baseline cTnI result (odds ratio [OR] 3.3; 95% confidence interval [CI] 1.7 to 6.3). Among patients with UA and no history or clinical evidence of HF at presentation, elevated levels of BNP were associated with a threefold higher risk of death through six months (OR 3.4; 95% CI 1.2 to 9.7) after adjusting for each these clinical predictors. Moreover, the baseline level of BNP remained predictive of mortality after the addition of peak CK-MB (as a measure of infarct size) to the clinical model (OR 3.3, 95% CI 1.8 to 6.3).
When BNP was evaluated as a semi-continuous variable, mortality risk increased in a graded fashion with rising levels of BNP (Fig. 2). As reflected by the respective chi-square statistics and ORs, the predictive capacity of baseline measurement of BNP in this population was maximal between 80 and 120 pg/ml, with a pattern of increasing mortality risk becoming evident at a concentration of BNP >80 pg/ml (Fig. 2). Although the chi-square statistic was slightly higher at a threshold of 100 pg/ml, this difference in discriminatory capacity did not achieve statistical significance compared with 80 pg/ml (likelihood ratio test chi-square = 1.84, p = 0.9). Five percent (n = 83) of the population had BNP levels between 80 pg/ml and 100 pg/ml.
Increasing levels of BNP (by quartile) were associated with, at most, a weak trend toward higher risk of recurrent ischemic events (Table 2). At a decision limit of 80 pg/ml, BNP was not predictive of an increased risk for new/recurrent MI (5.3% vs. 5.2%, p = 1.0) or rehospitalization for ACS (13.4% vs. 12.2%, p = 0.6) through six months of follow-up. In contrast, patients with elevated BNP at presentation had a nearly sixfold higher risk of developing new or worsening CHF by 30 days (5.9% vs. 1.0%, p < 0.0001), including those with UA (2.9% vs. 0.1%, p < 0.0001). The predictive association between baseline levels of BNP and CHF persisted at six months (9.1% vs. 1.8%, p < 0.0001) and was independent of the patient’s age, history of HF, ST deviation, or baseline troponin result (OR 3.0; 95% CI 1.6 to 5.7). As such, a baseline level of BNP >80 pg/ml was a strong, independent predictor of death or CHF at six months after presentation (16.3% vs. 3.6%, adjusted OR 3.2; 95% CI 2.0 to 5.1, Fig. 1B).
2.2 Stratification by BNP and troponin results
Consistent with previous work in TACTICS-TIMI 18 (16), an elevated level of cTnI at presentation was an important predictor of death (3.8% vs. 1.8%, p = 0.02) and MI (6.4% vs. 3.2%, p = 0.004) by six months in the subset of patients with data on BNP. B-type natriuretic peptide measurements added significant prognostic information to cTnI, discriminating patients at >4-fold higher mortality risk among those with both negative (OR 6.9; 95% CI 1.9 to 25.8) and positive (OR 4.1; 95% CI 1.9 to 9.0) baseline cTnI results (Fig. 3, left). Specifically, among patients with cTnI <0.1 ng/ml, those with a BNP level >80 pg/ml were at significantly higher risk of death through 30 days (4.5% vs. 0.7%, p = 0.004) and six months (7.5% vs. 1.2%, p = 0.0003). B-type natriuretic peptide results (dichotomized at 80 pg/ml) did not substantially alter the risk estimates for recurrent MI among patients either with (4.6% vs. 5.3%, p = 0.6) or without (3.0% vs. 1.7%, p = 0.4) elevated levels of cTnI. Thus, use of BNP and cTnI in combination enabled identification of patients with negative BNP or troponin who were at increased risk of death or MI (Fig. 3, right). Specifically, among patients with a negative cTnI who had elevated BNP, the risk of death or MI was 7.5%; whereas in those with a negative BNP detected as being at higher risk only by cTnI, the risk of death or MI through 30 days was 6.4%. Patients with both a negative BNP and cTnI were at very low risk of death (0.7%) or of death or MI (2.0%).
B-type natriuretic peptide also added important information to cTnI for predicting the risk of CHF (Fig. 4). In contrast with assessment of the risk for recurrent MI, patients with elevated BNP were at significantly higher risk of CHF and death or CHF, regardless of troponin status. Although in multivariable analysis cTnI contributed independent information with respect to the risk of CHF (OR 3.4; 95% CI 1.2 to 10.2) and of death or CHF (OR 2.1; 95% CI 1.04 to 4.20), BNP tended toward a stronger association with each of these end points (OR 3.9, 95% CI 1.7 to 9.1 for CHF and OR 3.3, 95% CI 1.8 to 6.0 for death or CHF).
2.3 Effect of management strategy
Consistent with the primary results of TACTICS-TIMI 18, among the subset of patients with BNP data, the invasive management strategy conferred a 30% relative reduction in the risk of death or MI (OR 0.70; 95% CI 0.46 to 1.04) and a 23% reduction in the risk of the primary composite endpoint of death, MI, or rehospitalization for ACS at six months (OR 0.77; 95% CI 0.60 to 1.0) compared with the conservative strategy. When results were stratified by baseline BNP, no appreciable difference was observed in the effect of invasive versus conservative management among patients with higher levels of BNP (pinteraction≥ 0.6, Table 3). Stratification by both cTnI and BNP revealed a pattern of reduced risk of death or recurrent ischemic events with the invasive strategy among those with elevated cTnI, regardless of BNP status (Fig. 5).
As a putative measure of the hemodynamic consequences of acute myocardial ischemia and infarction, BNP is a biomarker that has recently shown promise for risk assessment among patients with non–ST-elevation ACS (12,17,18). We now have confirmed prospectively our initial retrospective observations that BNP is a novel marker of increased risk of death and CHF in patients presenting with UA and NSTEMI (with or without CHF). In addition, our data provide new insights into the prognostic relationships between BNP and individual clinical outcomes that will be valuable in guiding clinical use of this marker. Specifically, we have demonstrated a strong association between the plasma concentration of BNP at presentation and subsequent mortality or new CHF in an independent population of patients with UA and NSTEMI. Moreover, we have demonstrated that our prospectively defined decision limit of 80 pg/ml with this assay appears to be a threshold above which mortality risk begins to increase in ACS. This study also reveals variation in the relative predictive capacity of BNP and troponin with respect to individual clinical end points, with BNP exhibiting a stronger relationship with death and CHF, and cardiac troponin being predictive of death and recurrent ischemic events. It is likely that this difference accounts in part for differences in the ability of these markers to predict response to a specific therapy, such as an early invasive management strategy. When used together in a combined strategy, these two markers provide a more effective tool for identifying patients at increased risk for clinically important recurrent cardiac events related to both ischemia and HF. Such information is likely to enhance our ability to appropriately triage higher-risk patients and more reliably identify low-risk patients who may be candidates for less intensive evaluation and therapy.
At this time, the implications of elevated BNP for decisions regarding therapy in ACS are not as clear. We found no interaction between BNP and the effect of the early invasive management strategy. These results are not entirely unexpected when one considers that BNP appears to be a stronger predictor of mortality and HF than of recurrent non-fatal ischemic events and that the benefit of the early invasive strategy in this study accrued primarily from a reduction in recurrent ischemic events. These findings contrast with those for cardiac troponin, which is a strong predictor of benefit from early invasive management, and provide direct evidence that different biomarkers carry different implications for therapy (13,16). Further investigation is needed to identify therapies that may favorably modify the risk associated with increased levels of BNP in ACS.
Additional research is also needed to elucidate the pathobiologic connections between elevated levels of BNP and the associated higher mortality risk observed in patients with non–ST-elevation ACS. We may, however, speculate as to potential mechanisms (19). Data from both experimental and clinical studies suggest that the plasma level of BNP may reflect the size or severity of the ischemic insult, even in the absence of myocardial necrosis. Specifically, in human models of ischemia, BNP levels have been shown to increase transiently both after exercise in patients with stable coronary disease (20)and after uncomplicated coronary angioplasty, despite stable intracardiac filling pressures (21,22). Others have shown the rise in the level of BNP to correlate with the size of an ischemic territory during nuclear stress imaging (23). B-type natriuretic peptide also appears to provide a very sensitive tool for detecting early CHF (24)and may serve to identify patients with unsuspected adverse hemodynamic consequences of coronary ischemia. As such, the available evidence supports a plausible relationship between BNP and the extent of the ischemic territory as a possible link to the increased risk of CHF and fatal complications. It is interesting that there does not appear to be a stronger association with future ischemic events or a more profound benefit from revascularization.
3.1 Study limitations
The results of this study address the use of BNP for risk assessment in patients with a high clinical probability of ACS. Additional investigation will be necessary to explore the utility of BNP in a broad spectrum of patients presenting with chest pain and low-to-intermediate probability of ACS. B-type natriuretic peptide levels may be elevated in a number of other cardiac and some non-cardiac conditions (25)and, as with cardiac troponin, must be integrated with findings from the history and physical examination to arrive at an overall diagnostic and prognostic assessment. Nevertheless, data from Jernberg and colleagues (18)using N-terminal pro-BNP for risk stratification among a less-selected population of patients admitted to a coronary care unit support the utility of measuring BNP and/or its pro-hormone fragments in the general population with suspected ACS.
This study provides support for interpretation of BNP as a dichotomous result at 80 pg/ml. However, given a stepwise relationship between increasing levels of BNP and mortality risk, the absolute plasma concentration of BNP carries additional information with respect to the magnitude of risk that should be considered by the clinician. Nevertheless, for convenient clinical use, a decision limit of 80 pg/ml provides robust predictive capacity with respect to mortality and CHF and is now supported by two independent data sets. This threshold is also qualitatively very similar to that established for the diagnosis of CHF (100 pg/ml) in a recent prospective study (24).
B-type natriuretic peptide levels change over time after presentation, and the association with clinical risk may vary based on the time of ascertainment. It would also be interesting to have data regarding the serial changes in BNP concentration among patients presenting with ischemia without necrosis. Only a baseline sample was obtained for core lab analysis in TACTICS-TIMI 18, and thus we cannot comment on the optimal timing of BNP measurement or kinetics of this marker on the basis of our data. However, we have extended beyond previous findings in which BNP was measured at one to three days after presentation to now demonstrate that elevated levels of BNP at presentation are also associated with adverse outcomes. It is possible that serial measurements may improve the prognostic performance of BNP.
An elevated plasma concentration of BNP (>80 pg/ml) at presentation in patients with UA and NSTEMI predicts higher short- and long-term mortality as well as new-onset CHF. B-type natriuretic peptide adds substantially to risk assessment for mortality based on cardiac troponin alone, but it did not enhance selection of appropriate candidates for early invasive evaluation in this study. These findings may be used to guide clinical use of BNP for risk assessment and triage in patients with suspected ACS. Additional work is needed to identify specific therapies that may reduce the risk associated with increased BNP.
The TACTICS-TIMI 18 study was sponsored by Merck & Co. Biosite Incorporated (San Diego, California) performed testing for B-type natriuretic peptide. Gottlieb C. Friesinger, II, MD, acted as Guest Editor for this paper.
- acute coronary syndrome(s)
- B-type natriuretic peptide
- congestive heart failure
- confidence interval
- creatine kinase-MB
- cardiac troponin I
- heart failure
- myocardial infarction
- non–ST-elevation myocardial infarction
- odds ratio
- Thrombolysis In Myocardial Infarction
- unstable angina
- Received October 27, 2002.
- Revision received December 4, 2002.
- Accepted December 26, 2002.
- American College of Cardiology Foundation
- Wiese S.,
- Breyer T.,
- Dragu A.,
- et al.
- Yasue H.,
- Yoshimura M.,
- Sumida H.,
- et al.
- Yoshimura M.,
- Yasue H.,
- Okumura K.,
- et al.
- Omland T.,
- Aakvaag A.,
- Vik-Mo H.
- Dao Q.,
- Krishnaswamy P.,
- Kazanegra R.,
- et al.
- Maisel A.
- Omland T.,
- Aakvaag A.,
- Bonarjee V.V.,
- et al.
- Hall C.,
- Cannon C.P.,
- Forman S.,
- Braunwald E.
- Arakawa N.,
- Nakamura M.,
- Aoki H.,
- Hiramori K.
- Richards A.M.,
- Nicholls M.G.,
- Yandle T.G.,
- et al.
- Morrow D.A.,
- Rifai N.,
- Tanasijevic M.J.,
- Wybenga D.R.,
- de Lemos J.A.,
- Antman E.M.
- Morrow D.A.,
- Cannon C.P.,
- Rifai N.,
- et al.
- Jernberg T.,
- Stridsberg M.,
- Venge P.,
- Lindahl B.
- ↵de Lemos JA, Morrow DA. B-type natriuretic peptide measurement in acute coronary syndromes: ready for clinical application? Circulation 2002;106. In Press
- Sabatine M.S.,
- Morrow D.A.,
- de Lemos J.A.,
- et al.